An alternative statistic for identifying lines useful for improving parents of an elite single cross

Theory and methods for identifying inbred lines (I w) with favorable dominant alleles not present in an elite single cross (I 1×I 2) have been developed recently. Selected I w lines can be crossed to I 1 or to I 2 to transfer new favorable alleles to the single cross. However, favorable alleles already present in the single cross may be lost during selection. It is important to consider both potential gain of favorable alleles from I w and loss of favorable alleles already present in I 1×I 2. The "net improvement" statistic (NI)=maximum [(I 1×I w-I 1xI 2)/2, (I 2×I w-I 1×I 2)/2] estimates the number of loci where favorable alleles can be gained minus the number of loci where favorable alleles can be lost in the single cross. Because I 1×I 2 is constant in an experiment, the method reduces to choosing I w lines with the best mean performance in combination with either I 1 or I 1. NI was compared to estimators previously proposed for identifying lines, namely: (1) minimally biased estimates (μG') of favorable dominant alleles present in I w but not in I 1 and I 2; (2) minimum estimate of an upper bound (UBND) on μG; and (3) predicted three-way cross (PTC) performance. Based on a set of maize (Zea mays L.) grain yield data, correlations among the four estimators were relatively high, but indicated that rankings of I w lines vary with the particular estimator used. Rankings of three I w lines based on the frequency of F2 test crosses superior to B73×Mo17 were identical to rankings based on NI, but differed from rankings based on μG', PTC, and UBND. NI also was the best predictor of the mean of the upper 10% ({ovx}0.1) of (I 2×I w) F2×I 1 or (I 1×I w) F2×I 2 test crosses based on simulated data. Being a simple statistic highly correlated to {ovx}0.1, NI may be useful in applied breeding programs.